Content users have a large variety of content options at their disposal. A user often has hundreds of channels or services available to watch at any given time, and many additional items of content that the user can choose to watch. Sifting through this content can be difficult. Some systems collect feedback from users in an effort to identify desirable content so that recommendations can be made. Such feedback collection systems, however, are rendered less effective when users do not offer their feedback. Accordingly, there remains a need to improve content recommendation systems.